Lag multiple, non-consecutive missing variables based on a prior one

[akmm10]

I have a wide dataset (could transpose to long, then be re-transposed if necessary) that has various outcomes at 15 time points. If missing, it will have "NA." I want to impute the first missing "NA" right after an observation. See example:

HAVE:

id	time1	time2	time3	time4	time5
A	1	NA	2	NA	NA
B	2	3	NA	3	NA
C	2	NA	NA	NA	4
D	1	2	2	NA	NA

WANT:

id	time1	time2	time3	time4	time5
A	1	1	2	2	NA
B	2	3	3	3	3
C	2	2	NA	NA	4
D	1	2	2	2	NA

Any help would be greatly appreciated!

[Tom · Accepted solution]

Just work from right to left.

data have;
  input id $ (time1-time5) (:$2.);
cards;
A 1 NA 2 NA NA
B 2 3 NA 3 NA
C 2 NA NA NA 4
D 1 2 2 NA NA
;

data want;
  set have;
  array t time1-time5;
  do i=dim(t) to 2 by -1;
    if t[i]='NA' then t[i]=t[i-1];
  end;
  drop i;
run;

Result

Although I do not understand why you would want those NA strings in your TIME variables when the other values look like numbers.  If instead of that constant text you set the values MISSING then you could use MISSING() function instead of the equality operator in the IF condition.

if missing(T[i]) then ...

[akmm10]

Oh, simpler than I thought! Thanks!